Central birdie:There will often be a large birdie in the central channel immediately after changing config. Stop/Start often helps with this.
Flagging: Whenever you select EDIT in the correlator, the flagging is reset so, you will need to redo the flagging.
Rear Transfer Module (rtm) Problems: If you have some baselines with spikes, but no power (in the amplitude plot in spd), you may need to initialize a Rear Transfer Modules. Check the “Data Links” tab in the cacor gui. Modules that need to be initialised will generally have red text and large numbers (>100000) in the 1ms column. To reinitialize:
cacor> rtminit ca0# freq pol (e.g. cacor> rtminit ca03 2b).
If this fails, it is sometimes necessary to reset the digitiser itself:
cacor> diginit ca0# freq pol (e.g. cacor> diginit ca03 2b).
Lost Delay Calibration: It seems to be possible to drive the delay calibration so far away from the correct value that you start calibrating on something(?) else. The first symptom is often noisy data (or high Tsys). Solution is to reset delays (cacor> reset delays) and restart calibration.
Amplitude Glitches: Check for amplitude “glitches” in vis. Reprogram suspicious IFs.
Memory Problems (mostly in pulsar configs): If every second cycle is different (e.g. the spd amplitude plot looks good one cycle, then RS the next (then good the next.....), you may have a memory problem. To solve this:
stop caobs
caobs> stop
then rest the correlator memory
caobs> reset mem
This is only recommended for pulsar modes.
Commands in High Time Resolution Configs:
Cisleep: In high time resolution modes with many channels(>100), it may take up to 2 seconds at the start of the cycle to transfer the data. This data transfer interferes with setting the fringe rotation, and therefore the fringe rotation needs to be deferred. 4 seconds should be sufficient.
cacor> cisl[eep] 4 sets the delay to 4 seconds
cacor> cisl[eep] returns the current value
Archiving High Time Resolution Data: The high time resolution data will only be saved to disk if carhtr is turned on. It is turned on by default, but in observations with very fast sample rates, it should be turned off when the high data rate is not required
cacor> carhtr on | off turns the archiving on or off
cacor> carhtr returns current setting
“Big Humps”: If an autocorrelation spectrum has a hump (~105) in the amplitude around channel 1800, the datalink block for that input needs to be reprogrammed
Determine which block needs reprogramming from the CABB datalinks tab -
select the EDIT button
reprogram the bus calibration for the block -
> prog # cal
restart the scan and inspect the data (again) in spd. It takes a few cycles before the spd display updates.
In everything, always remember to check both frequencies.